arr

Most literature on this topic deals

with propositional modal
logic
We will be both
exploring propositional
as well as modal
first order logic

Classical order issues like

first
functions and
symbols have

treatment Moda
Straightforward formal
the these
logic requires rethinking of
things
ut propositionalmodallogic

formalisation is a fundamental tool

for clarifying language leading to better

understanding of thoughts expressed through

language
Formalisation involves abstraction and
idealisation

The word model is

key A

mathematical model the universe

of
is not the universe It is an

aid to understanding an
approximately
correct though partial description the
of
Universe

We will present a mathematical mode

the modal We
of aspects of language
formal techniques to clarify the
apply
interactions of modal issues with other

important notions such as

identity names

definite descriptions
and
In order to present a
logic formally
we need several items

We need a
rigorously defined language
in its pure uninterpreted syntax We

also need a semantics to

give
to the formulas
meaning of a
language
which will let us out those
single
formulas that we are interested in

i e the valid ones we want a

notion
of proof to write down

compact verifications that certain

valid must
formulas are we

connect notion
finally the
of proof
with the semantics can be
so we

sure it is exactly the valid

formulas that have proofs

 In classical propositional logic
truth tables provide the semantics

Valid formulas are those which every

line evaluates to T such formulas
are called
tautologies
what is a modal

A modal the truth

qualifies of a
judgemen

Necessarily 31 the
Possibly are

most important best known

modal
qualifiers

They are called diethic modalities

from the Greek word truth
for
In trad
terminology necessarily P

is an apodeictic judgement and

 possibly P is a problematic
judgement 2
whereas by itself
is an
assertoric
judgement

The most widely discussed modals

Other than ahethic one are

modalities which
the temporal
deal with past present
future
will be was has been

will have been the deontic

modalities which deal with

and
obligation permission may

can must etc and the

epistemic doxastic modalities whic

with knowledge and

deal
 belief certainly probably perhaps

etc
surely

The modals mood and

capture tense

aspect which can all be lumped together

as adverbials i.e the modifiers
and derivatively
ofnouns
adj or verbs

e.g John is happy

be
the blank space com
filled by
any of the modals alethic temporal deontic

epistemic doxastic

Alternatively we can insert modal

into
qualities
It is true that John is happy
 This is modifying the truth
regarding
of
the claim

Hence 2 characterisations that are

equivalent

the modal qualifies the predicate

the modal qualifies the truth a claim
of

Note that not all models

give
to modal there
rise
logics Usually
must be two modal operators that

conform to particular logical principle

Can there be a modal

logic
Aristotle set the tone for
trad which
older
of logic
stretched to Leibniz by including
the modality in his
stray of
book on the
syllogism
The modal never attained
syllogism
the new universal
acceptance of
Aris more familiar logic of
statements
categorical

modality The
People opposed
Primary Claim was that modality
the content
had no
place
in of
a so the modal is
judgement
inert w to the
regard logical
connections between that judgement
and others
 content
Conceptual is the part

of judgements which
affects he
possible inferences

Compare the Greeks defeated the Persians

and the Persians were

defeated by
the Greeks Both Get A pas
forms of
a
judgement Frege 1879

the
argues
have exactly same

connections any judgement

inferential
entailed by the first is also entailed

the second
by

Frege says the content represented by

f remains the same whether preceded

by necessarily or possibly or

at all But it does not

nothing
that the apodeictic problematic
follow
or assertoric judgements all have the

same
concertmaster contrary
to what he says modal

distinctions de affect possible

inferences

D it is that
necessary P
P it is possible that

Now P D P i e it's actual

so it's possible is considered

usually
to be kid Its converse
 4232 is not
note P SQ read if 2 then Q

The most
effective way of showing
can be a modal
there logic is

to construct one

What are the formulas

We must be more
specific about
we
me formal language are
using The lang

of propositional modal extends

logic
that Clas
logic
of pnp Classically
are built up from
formulas propositional
letters variables P Q 12 Stand
They
for unanalysed propositions Sometimes
T and t
we may use for the
perse
These are called propositional constants

Prop letters are combined by prop

connectives 7 is unary and a V

J I are binary

D necessarily I possibly These are

also unary

Definition pre modal formulas

The set of pop modal i's

formulas
specified by the following rules

1
Every prop letter PQ is a
formula
2 X is
If
a
formula so is X

3
If X Y are formulas and o
is
 a binary connective X Y is
a formula

4 X is a
If formula so are DX X

e
g
Der Q 246191 is a
formia
and is read I
if is
necessary
and Q is possible then it is possible
that both P and Q

Some texts L M D D
use
for

J is oft used in place of D

is a
person't knowlege is concerned

Ha Ky etc can be used

if
more than one person's knowledge
 is concerned a multi modal
logic

Likewise modal operators

corresponding
to actions can be considered e
g
the action be that
might of going from
NYC to Fran
San
If D is intended

to denote this action then

it
Pass

me drop it is 12m and Q is

it is 2pm P DQ would

read if it is 1pm then I

after
go from NYC to San Fran it will

l
be 2 pm

This also falls in multi model

logic
Aristotle's Modal Square

A minimal
req for modal logic is

that D and D Satisfy the Modal

square of Opposition below This is
worked the
due to Aris who
out

basic logical connections between

necessity 31 possibility in his De Interactatione

The negation of IP is it is not

possible that P
The negation D is it is
of It
necessary that P
further
It is possible that

f It is not necessary that note

 49 71777

13 747
here we
get m so

It is necessary that
It is
It isnotpossiblethatnot n not possible that
p
y

It isnot necessarythatnot
It is possible that It is possible that not P

The Ms O is analogous to the square of opposition

G No f are G
All f ane
p

some f are G Some f are not G

The top row are contraries cannot both be

true

The bottom row are sub contraries

cannot both be false

The columns are sub alternative

top bottom

The diagonals are contradictories cannot

have the same truth valve

In formal Interpretations
we earlier characterised a modal as
an adverbial that
fills the blank in sort
like

John is
happy
It is however
actually more
complicated
Adverbials in their
differ favoured placement

Necessarily must precede the adjective

now then can occur either before or

it here
after there must come
after
it

This is
independent of the
underlying
logical sense It will be helpful to

Stardise our
frame like

John is happy
with the modifier coming after the
adj
The modal we insert
qualifies John's
happiness and
opens a
spectrum in which
modification his happiness
of can be compared
 we can extend this as

John is happy under interpretation i

Think
of an
intern i as a
particular
assignment of truth values to the non

logical particles in the

proposition with
the particles understood the usual
logical as

boolean
functions

On the simp reacting of D and G

a
proposition is
necessarily true if it
comes out true
for every interpretation
i e
if it is a
logical truth a ch a

proposition is possibly true

if it
comes
Out true for at least one interpretation
i e it is
if consistent

we then have
 DJohn is happy
H John is
happy

to
corresponding
ti John is happy under
interp i

Ii John is happy under C

into

John is happy without a modal operator

is understood as John is happy under

interp io where io is the interpretation

to the real world
corresponding

It is more usual in mod

log to
speak
of a
possible interpretation as a possible
world state These notions such as

necessary truth E true in all possible worlds

 is derived from Leibniz

how this works

for DR 2
Observing
if P is true in every possible world then

it is true in the actual world

I D on the other hand is invalid

because because I is true in the

just
actual world it need not be true
in
every possible world

Modern mod log

employs a nodi
of
Leibniz's notation into
by taking
consideration the
way modals relate to
possible worlds This is called an

accessibility relation
temporal
worlds are ordered
Eg
by their
place in time
Take a world D to be accessible from
 world f if f occurs no later in the
etimecsl
temporal order than A.ie ifIgtig is

necessmiythe if it is true at

every accessible possible word and is

tee if it is true at some

accessible possible word

Let wi be a
possible word and we be
world
the actual
JI is the
accessibility
relation x
Ky is read
y is accessible
from X

Then D John is happy

corresponds to

Iww John is happy at wi

for all wi in the relation

I to Wo Cine for all

Wi accessible from Wo

Consider DROP again

This says that I i's
if the at every
possible accessible world then it is true at
the real world To remain valid the real
must to
world be accessible
itself which

it is according to the temporal relation

given

If we
define the real world as not

accessible to itself this formula

is not valid

Kripke 1963 a introduced the notion

of accessibility and this notion makes

rich
it possible to model a
variety of modal

logics

Let us see
us to extend our
informally now
J2 enables
understanding of
modal operators
 Consider P DD LYE 61.3 Let AP mean
it is conceivable that P Then the above

says that if there is a

possible world we can
Conceive of from the actual one i e an

accessible conceivable world in which P is

true then from every accessible conceivable world

there is an accessible conceivable world in which

P is true Is this so

If 2 is conceivable is it inconceivable
that it is inconceivable that

consider the following

example
it is reasonable to suppose
disciple of a

Euclid Dq would be of the perspective that

it is inconceivable that parallel lines meet
but in contemporary mathematics this is
not inconceivable to us A So
from A's
it is conceivable that
perp parallel lines
1
meet making the antecedent of 3 true
but there is at least one conceivable situatio
 if A looked in De's perp from which

it is inconceivable that parallel lines meet

making the consequent 13
of false
Presumably we A can look into
De's
perp but not the converse because
A knows suit about the past but
DE knows about the
nothing future
I in the case
of conceivability ye
is not and C1
symmetric 37 does
not hold

Once accessibility among world's has been

highlighted as the preeminent characteristic
of
mod can
logs we
interp mod
operators
far from their originally intended meaning s

An example of this is Harel's lase

dynamic logic

One can generalise the subj of modal

into attempt to characterise
logic an talk
transitions or
alternatives
of deformations
or
What are the Models

Possible world semantics have

become the standard introduced by
Kripke and independently
by Mini tikka 1961

A valid formula i.e truth

logical of
Classical prop logic is one that comes
out true for every possible
assignment
truth values to the statement
of letters

C P Q Tif
g TT Cf F F T formula
always holds up w all these

Difficulties found in interpis of classical

logic
is exacerbated in mod since we
log say
There is a
possible world such that

and so we
quantify over possible worlds

criterion
According to a
of ontological
commitment by Quine 19481 this means we
 are committed to the existence such things
of
But some
of here are intended to be
mere possibilities not actualities and so
we appear to be more committed to

the existence of things we would not otherwise

want to be committed to It looks as
though
we have increased our store
of existets Not
only do we have the
way things are
but also the
way things might be The
same
difficulty is present in the classical
case for we speak of multiple interpretations
And we consider
all possibilities
an truth table lines and these poss Cannot
all hold simultaneously The problem is then
not with the notion of possible worlds but
abt he philo theories abt hen

Definition frames
A frame consists not set whose
empty
a
of
members are
generally
and sin net Kong edgatssjb.gr h
accessibility relation i A frame is the

pair 2,12
whilst 2055 world is
suggestive terminology
it commits us to nothing TRD means D is
acc from P or D is an alternat eword

to P

Definition Model
into model
A frame is turned a model
by
which Letters are true at
specifying prop
worlds mode
which A prop modal model
short
for is a triple 1,12 11
7
whee J R is a
frame and It is a

relation betwe possible worlds and

prop Letters

If 511 3 2 is true at T
If TIP
then R is at 17
false
Incidentally It is
often read as forces In
class each line truth table is
log of a a
model And a truth taste line is specified
 letters
completely by saying which prop are

assigned T or F Modal models are more

complex in that
family of such truth
a

assignments is involved one for each TED and

there is some rel beth these
assignments 32
Now in the classical case once truth
values have been specified at the prop
letter level they can be calculated for
more
complicated formulas

The modal analog of this is to calculate

the truth valve
of non atomic formulas
at The
each possible world rotational
mechanism we use is to extend the
relation beyond the prop letter
I level

XH X forces Y finite forcing rearm

DCE X A i's
formula
Iit A x satisfies A A is satisfied inx
 Definition Truth in a model

let 9,2 it Yet ice

It is extended to arbitrary formulas as follows
For each
Gg
C Sit X
THX
2
THAY E THX i
MAY
3 MH DX EVIE g TED DAX
4 MH
AXEFAEJ.SK NDHX
The first 2 clauses
say that at each
world the
prop r and n behave in the
usual truth table way The last 2
however
are what is special

3 says necessary truth is equiv to truth in all

possible worlds but limited to those that are accessible
from the one you are dealing with
 in
y says possible truth is
equiv to truth

some accessible possible world

Suppose we think V
of as
defined
in the usual way X V Y T GX r Y

Then the various conditions above

using
i e t 4

HSH Xv Y E Tt Ger Y
r axe Y
TACKY F Xu
E MI XVI Y
THX U THY

THX Y TH Xi Y

SIX TH Y
f LHSfalse the Rees is true MAX V M If Y
ritxority
 THEY E THX EMILY

THX notation does not mention the

model only a
possible world and a

formula This is fine if

it is clear

which model we are with but

working
more than one model is involved
if
we write M THX where 4 8811 7
model We take this to mean
is a

is true at the world T of

x

the collection I of possible worlds

of the model M

exercise

I MH DX TE X

TH DX I DX
 MH IX D x

I TH axes x

TH DX I DX a TH AXEL X

world
x is necessary at
Tiff X is

necessary r
X is
possible at world

I X is possible where T is
iff
a possible word

Examples of Modal Models

models will be diagrams

given using
with boxes
representing possible worlds
for 2 worlds with such a model S and

D PID we indicate this

if by
drawing an arrow from P to D
We say explicitly which
Roy letters

Are the at particular worlds If they

are not shown at at a word
they are false
 certain
We begin by showing
are not true at worlds
formulas
in model then
a
particular
move on to
showing certain other
formulas aretes at worlds
of
models under very broad assumptions
abt hem

Ex 1

Elite Irina
here DH Rvd All P and

R It 2 VQ a HQ o

D and so are the

only possible
worlds accessible from M and PNQ
is true at both them TH D Pva
of
Hae 2,5122 kava de ED or
On the other hand we do not have

711 02 as that would imply since Tylor

we
would have dit aspers which is false

also do not have Mil DQ as

We

if we did this would imply DILQ

i
ND
i we do not have THDIVTH DQ

i Deva 0 Dev DQ is not true

at M and neither is

DP a
4070 na

x2

TEM
This time JP P

5H 2 DII P Δ
is the world accessible
only
from M we had MADIR
If
it would follow that 11
172
and 2172 which

as we can is not
see

the case in this model

S PS
 EX
TIM
111 49 D is ace from M A DM

111 1 IT would
imply DNA Not
which would
imply there is another Only the
world actual
accessible D from which is world
Untrue under possible world semantics Wo is
access g
Da 42 E FD'eg ARD'd D'll to
its lf
But THE Js 4721
THAT ODD

IIa's actual
Iifa
Eka a when Q
Ip it's possible
DR it's necessary
Ext
This is counterexample to lots
of
a

interesting formulas

II

me
f
p is Awe at D and D is

world D
the only act
from
then II is true at D Since

D2 is true at D and D is all

the worlds acc from D DIP is

true at D So is D DDP and
So on
 Now since D is all the worlds

T also DR is true
ace
from
at 7 But since DP is also

true at D then DD is true at

T and so on On the other

hand I is not true

itself
at T Thus interestingly
all the are
following formulas
false
DR 3 DDP 3 DDD R
etc
which is to the
contrary
case
general
 Now we turn to examples showing certain formulas
be true at worlds under
must very general
circumstances

Earlier we showed 1 Eva D OR VAQ

could be under the
false right circumstances
so we cannot
assume IT distributes over
Now we show I does distribute over

n so D 219 DRA Da is true at

every possible world model

of every

We will not be looking at diagrams since we

re with the class all models

dealing of

5ps 2,2211 7 is a model and TED

Also Mil I Paa we show it
follows
sps
that MH DR a DQ
Let D be an
arbitrary member
of J s t

1121 TH I Pua it follows DH 2nd

Din a n Dita i D was
arbitrary
at
Fossick accessible from T

both P and Q are true hence at M both

32 and DQ are true I TH DP a DQ

In a similar way Bar Da DD Rr Q

be shown is true at world
can every possible
as is D 20070 DRO DQ

looking at the first one

if g DD is
arbitrary and THI n DQ

THI n THD Q

Alter D HQ
Alt Ra Q It is distr over n

TH Deana
as
for the second

J 3D is
arbitrary and 711 1312 Q

D it S Q

A 11 p p D a Q it is distr overs

THI S TH DQ

For the converse

Tin DR J TH DQ

D it 0 Bit Q

D it 23 Q It is distr over D

E T it D 209

sometimes
binary relations
satisfy
special conditions and many
have standard names
of these
A relation J2 on
J is reflexive
A
f Meg MRT holds that is

if every word is accessible from

tself

R is transitive 7215 11282

if
Tros

R is
symmetric if 721 12217

Sps 4,32 11 7 is a model in which

Then TH DP
12 is transitive
01,17
Sho
Tir DD
Sps TH D We mush
 and to do this let D be member
any
of J
s t
MRD and show DAD

and to show this let d be member

any
2 s t DRI and show Sharp But
of
is transitive so M It or on p

Dlt Dp 511 1319

Similarly

1 if 12 is reflexive
DR 2 treat an

members
of 2

2 If 12 is symmetric 20542 is true

at every member
of J
3 92 is both and transitive
If symmetric
 then I 231342 true at an arm
g
g

at l
ooking

12 reflexive MT Meg
Sps T HDR
it Rt THR
P It DR D P

oohing at 2

R is symmetric TIDE 1125

Ses TH R DH DP
Meta
DRT at It
1 D 11 52 gym is
T It Be
looking at 3

wts f RODA
5ps THEY
TRD ndra inpr

f
mookt MY
need Dita routes

if 711479 Alter suit

WWE assume Dit without assuming anything

If about a
2042
111 49
117th OH DIP DHAR IPO

E IF i AHA AAH EY
radar
I Be

P S Q means
if 711 3 then Mit Q
Note MitESQ

Some important logics

wide modal
There are a
variety of logics
because
diff intended uses and
partly of
because our intuitions about
partly
 and not
necessary truths differ or
fully
developed

whilst everyone probably agrees that necessary

truths have
necessary consequences opinions
differ about whether necessary truths are
51 or Few
by necessity contingently may
have thought
about whether a
prop whose

necessity is possibly necessary is necessarily

true or even possible

once iterated modalities arise intuition and

experience begin to be inadequate In part

diff formal mod logs have arisen in an attempt

to capture the differences in the behaviour of interated
modalities Variations between mod at
logs
the propositional level have minor significance

mod were characterised

Originally diff logs
by giving diff sets of axioms for them
 world
One the early successes of possible
of characterisation nosy
Semantics was the
of
standard mod logs by placing simple
maths conditions on
frames Conditions are

placed on frames Although models are what

we deal with most
often frames play a

central role

Definition L valid
We
say the model I 12,11 7 is based
the frame 28,72 A formula X is
valid in a model 8 12,11 7 if it is

true at
every world of Z A formula X

s valid in a
frame if it is valid in

every model based on that frame finally

L is X is
f a collection
of frames
Latif if X is valid in
every frame
X is valid in
n
meaning every model
based on every frame in 2
Definition frame conditions

let 2 R be a
frame
it is
reflexive if tied TRT
it is symmetric IT TRIED 125
if DEJ
it is transitive V7 D O
if EL
TRD ADRA TRA
it is serial
if HTELFAEZ
Sit TRD Surjectivity
it is linear
if UP DEJ either MD or

DRM
We now list some important modal
with the frame conditions
logics along
required them
 8
frame conditions
Logic
K no conditions

D serial

T reflexive

B reflexive symmetric

KY transitive

S4 reflexive transitive

55 reflexive Symmetric transit

 are
Different modal logics
characterised as
semantically
the L valid
formulas for particular
classes L frames
of
e
g T is characterised by
the class
of frames having
the that world
property
each

is accessible
from itself

From henceforth we will Use

T to denote both the class

of reflexive frames and the

determined by that
logic class

We saw the
previously that
relation Dana DDP a DQ is
true at world
every of every
model In other words it is
k valid
Similarly
1 PODER is 55 valid

If a
formula X is Ky valid
then it is true in all transitive
so in
frames and all
 transitive and
reflexive frames
it is Sy valid and so kill validalso
ka is a
subtogy of
54

Similarly 113 is a sub

logic of
55 and k is a sub
logic
all
of others
Finally every
reflexive relationisautomatically
seria
This is all summarised in
the
following diagram
 55
n n

B 54
n 7
r

D KY
7
r

K
Every frame in T is also
a
frame in D but the

converse is not true

Still it is conceivable that

aslogics D and T are

the same That is it is

conceivable thtformias
arevalidinthethoclasses
offanestowentsis
not the case here e for example
 A
sty
This is a D but 17909
frame
is not true here unlike T
frames
as D
diff
T are
logics

Logical Consequence
In classical the notion
logic of
follows from is
fundamental
Loosely speaking one a
says
formula X follows
from a set

of formulas S
provided X must

be true whenever the members

true
of
are
More formally one X
says is a

logical consequence of S in

classical
prop logic provided
that for every assignment truth
of
values to
prop letters
if every
member S evaluates to T then
of
X also evaluates to T This is
symbolised as
often SFX

for example the compactness

property says if Sk X then

for
some finite subset S.ES we have
also So tX Tarski instituted the

general study of consequence

relations and their
fundamental
role has been accepted ever

since Things are more

complex for
Modal
logics

Definition consequence

let L be one the

of frame
collections in the table
given
above Also let S and U be
sets X be
of formulas and let

We X
Say
a
single formula
is a in L S
consequence of
as
global and U as local assumption
and write
SEV X provided
for every frame 2,32 in the
collection L
for every model
22,12 11 7 based on this
in which all
frame members
of
S are valid and Urey at
which all members U art
of
true we have THX

Thus
Suzu X means X
is true at all worlds L

models where the

of
members
of
U ametre provided the mem
of
S are true throughout the model i e at
 world
every

local and
global assumptions
play
quite different roles

Eg
Take L to be kg the collection
all
of frames S O so there are

no
assumptions and U to
global only
contain DR 2 and X to be

DD P DR The
following model
shows
0Fk 132023 5 DDP DDP does
not hold ire DDP DDP is true at
all worlds where 1722 P is true with
no
assumptions for the collection all worlds
 FIN
here DH 2
2 83 31 3

TH DPE HR Consequently
TH P 711 D TR
TH D O so

the only member of 1202 is true at

M On the other hand Dls 132

there are no worlds alternate É D at which

p fails A is me
poss world ace
only
from 7 it
follows TH DDP 17127
TH DD 2172 We have a counterex

to PFK 912073 319520132 fails

Every global assumption is valid in the
model since there are none
But at T
all 1041 assumptions are true and 131330132 is
here
not DK MADE but 511 DD

AxVacuous Truth
Why is this the case Why is IT true in

a world when there is no world accessible

from it In Logic Language and meaning

Vol 2 24 it explains wheywon Dp
pg
since there are no worlds at all
which are accessible
from W so

that p is true in all the

of
non existent worlds which are

accessible

This seems to suggest that

if there is no

world access ble from M then I il DP is

always
true
 The here is vacuous truth 90 there is
key
no accessible world where
I can be false

it is vacuously tree in all accessible worlds

i e D

On the other hand there is no accessible

world where I can be true it is not true

in any accessible world Thus it is also
not
possible i e
743

This explanation should aid an understanding

to the same concept Classical
logic
in

i e think abt Axe as opposed to

Fx
EP
Comparison with the example shown
above will make clear the
diff
in behaviour beth local 31
global
assumptions

Ege
This time we show that

DR 23170 713132252 does

hold

Let CJ 22,11 7 be a k model in which

I 02 is valid in all worlds

global assumption
All local assumptions hold since there and none

We show TH DD DD

5ps T 11 DDP Let A be arbitrary and E3

holds
MR D Then 111 172 DROP everywhere
Du 0202 DH DP D 11 7 111 P
i
D was arbitrary 711 179

Many of the
general properties classical
of
consequence over to modal consequence
carry
for instance Classical cons obeys a

monotonicity condition X is
a
consequence a set
if
of it is also
a consequence of any extension that
set A similar
of
result holds
modally

Proposition monotonicity

5ps SEU X and SES n U E U

then S EU X
Proposition compactness

525 Sky there

X Then are
finite
sets S Es n V Eu sit S.EU _sx

one the most used

of features of classical
consequence is that it
obeys the

deduction theorem this States

Classically
SUEY EX I SECY X
It is
often used to
replace the
proof
of an implication Y ox
by a

derivation a
simpler formula X
of
from a
premise Y But mod cons

differs significantly here There are

2
versions
of the deduction than
 z
on whether Y is taken as a local or

global assumption

Proposition local ded

SEU U Y XESKU SCY x

global ded
Proposition
SUEY FLU SXESKUUEY.BY DAY SX

The definition modal

of deduction
differs significantly from book to
book

Definition
Abbreviate SED SX
by SE X
Exercise
snow that off DR 773 5117207 does
not hold but 172023 Fk DDP op
does hold

i 01 402033 51713222
no
global assumptions
need THD 2 DDP P

find a
counterexample

Ty THAT AT ADD AH DR Vacuously

bafTTHD
É so

If but also DROP is true at P

hence M DD DP if DROP is
only a localassumption
here DK MADE but 511 DD
if P then Q en
if 73 Q could still happen 27 07177 900 4609
but
if 20 and
I happens this is false stso It P Q
 ii 172023 DDP 23

DID true at every world

no local assumptions
Assume 711 DDP let D be
arbitrary
It DDI TH DEATH p truein

Dp i TH DD D th dit Da n guy
Dit Pa Dan

D P in all worlds
necessarily in D itself
aft from A but not

as a actuary in
Ditsey

Temporal Logic
We have
stronger intuitions about
temporal
worlds than we do about the more abstract
and austere
alethically possible worlds and
frequently we will
find ourselves the
using
emporal interpretation to aid in the
understanding
f modal concepts
For Frege a
complete a
thought
Gedanke is an eternal non
linguistic
object whose truth value never
changes

Quine has a
corresponding notion
of an
eternal sentence The sentence

John is
happy
s not an eternal sentence it does not
by itself express a complete
thought This
is because in
actuality there will be
periods
time
of where the proposition
expressed
by
his sentence is
false it is not
by
timelessly exclusively true or
itself
false
If we were to add a time determination

to the sentence we would stabilise its truth

valve
Let us
qualify where
John is
happy The sentence

frame is
effectively filled with
a time index t

John is happy at time t

where t
occupies a position that
can be bound
We
by a
quantifier
sps there to be a
range of
times at some
of which
cerhaps John
is happy and at others not 525
we have
 John is happy in Nov 1994 A

as true Now Sps sont happens

Some months later and we can

take

John is
happy in Feb 1995
fax
as We can
false existentially
generalize on the true a to the
get
true

Ft John is happy at time t

Given the HK have

falsity of
we

the also of its

falsity universal
generalisation

ft John is
happy attinet
Treating John is
happy as John as
at time t in
happy effect is to

that John is
regard happy not as a

proposition but as a
property times
of
On his reading John is happy at time t
is an
open sentence when t is inside
the scope of a
is
quantifier it

bound by the
quantifier when outside

the scope a
it acts as
of quantifier an

indexical now the designated

denoting
actual time

eg
Sky temp intern
D John is happy I ft John is happy at time t
John is happy I FG John is happy at time t
As with the alethic inters p f a holds
177 holds
and 4202 doesn't and Of
and 2 IT doesn't

Let us our interp with

complicate thing
these and past tense
operators
future
F case that p

23 Ijiggomthete
Cetme case that p

where the very in p is now untensed

It is common to
define

GP Itaillatwaysbeken Case that p

112 It h alwaysbente case that P

where G I 777 and tip P P

these tense operators we can

Using capture
a number of natural language tenses
Let 2 be John wins the election's Then
have the following inner natural
we
ways of
language tenses

a P John wins the election

John will win the election

b Ff
John the election
C RP won

d 22 John had won the election

e FP John will have won the election

John would win the election

f Pf

We two diff modal

can dev logics
from these operators by taking D to be GP and

42 to be Fl or
equally well DR is up
and 42 to be 22

These modal ops can be nested and so

introduced complications are nontrivial we have

not made provisions for iterations

of temporal
operators yet This is clear from our
quantification representation If He Says
there will be a time t such that

John wins the election at t then

putting another temporal Oa in

front win

appear to have no
effect since there is

no
free variable left to bind Yet from the

intuitive reading to the mod ops

we
give
above there are obvious distinctions

e is dist.fr andccisdist from

d How do we handle these cases

Let the present time Se t Then we have

for

F John wins the election

2 John wins the election

G John wins the election

H John wins the election

respectively
Ft t SE n John wins the election art

Ft t CE n John wins the election at t

Atl t E A John wins the due at t

ft t's t d John wins ele at t

i
Then and become
e
Cfl
e Ft Ct E ACTE E'at d John wins elec att's
FE E tA FE KE SE d John wins eve at t 7
f

for Cdl we have

d Ft Eat NCTE t at m John wins erec ate

t time
present
e says that there will be athe
ine
Informing itins the election we will

have won the election Cursent time

 ff that there is a in the past
says
time after
whicunjointiisteneetontidaintu
eation present
fine

This is our more elaborate temp interpretation of

the mod ops

We can mod logs by defining

get diff
and I and in these mod
D differently diff
statements or not
logs different may may
hold

we can take D to be G and D to be

e.g
mod
f This gives us another log
e we can take Beto be an Ga and Jp to be
g
another mod
V F This gives
us
log
There is an ordering relation that is imposed
by time in
temporal logic later than The

temporal ordering is transitive but not

reflsyfm so
if
we take DR to be Gd

a prop is nee true if it holds at

ordering If on the other hand we

an Ga then a
take De to be prop
is nee true
if it
holdsnowAnd
I 222 fails
forevertaf for the

former
but holds for the latte

Epistemic Logic
we can also inter the Mod to
symbols
create a
cogic of knowledge Hintikka 1962
intros the following D and D resp
readings
Nap a knows that P

It knows
Pap possible all a
is for
that P

This is a multimodal logic each

subscript identifies an individual There

similar vocab
is a
for belief

23 a believes that P

with
Cap It is compatible
a believes that P
everything
the usual
In each case we have

mod
interdefinability of ops
 Kar Jan
28 I Eat

The tums Hintikka include an

of
tautologies plus

41cal Q1 Harstad
Ka O

Ka OKakal
rules
and the two
of inference
modus Loners Y follows from X and

X OY and Necessitation
Fall follows
from X
a will be our k axiom
corresponding

to condition le in the table

frame
Likewise be
given will our

T axiom and will be our

4 axiom With T and 4 on

top of
K we have what is called

an 54
system

Kal is
perhaps better read as

It follows from what a knows

that 2 or perhaps as
a is entitled to know that P

for in this system one knows

 the
all
logical consequences of
one knows Ps Q
anything if
is logically
true and
of Kap
then Thad In this
logic
then the knower is treated

as
logically omniscient This

feature of the
system
is made

because
even more
puzzling of
S which is called the

Principle positive Introspection

of
one cannot know and
something yet
to know that knows
fail
one
that
something
Formula xxx is
justified despite
its title not
knowing that
because

I carrieswitta certain

mental which
feeling knows
one

one has
by introspection

Hintikka is
working with the view that

knowledge is
justified true
belief
So a is because
justified if
One knows ther f one must be

justified in
believing that P and
one cannot be without
justified
being aware of being justified It
is not just that if one knows
that P there are reasons

P rather
justifying if
one

knows that P these

justifying
reasons must be
your reasons
and so cannot but be
you help
aware
them
of Logical
Omniscience
that Q
Lifeless
if follows from R on
logical knower
grounds every
is aware
of these
grounds
This is a
very strong condition
On it
knowledge and is
why we

prefer reading as
 a is entitled to know that R
rather than a
as knows that P

course there
Of are some

constraints on
knowledge It is
interesting in this
regard to note
that by contrast with xxx

the
Principle
of Negative
Introspection

Ka O
Karta
is
not a theorem this
of
system Again this has nothing
to do with The
introspection
reasons are a bit more mundane
you not
might know that 2 and
yet believe that you do know
it Cat however
requires
that
if you don't know that P
then you know and so
believe
you don't know it The
also be
following fails to a

theorem

E
700ha
And this seems
appropriate for
a statement 2 be true
and might
yet one need not know
that with
it is compatible everything
One believes The all
falling
do hold in Hintikka's
system
IA
Kaku Ka
I I
Ka Pa
X
KAREL aka

Ia expresses the transmissibility

of Edge The same

does
get hold for belief
i.e hold
23,13 233,2 does not
 essentially tells us that every world has
an alternative which we know from the

reflexivity expressed in the Ka 0 axiom

It can also be shown that Jeep off Q

follows from PDQ This is a
form of
logical omniscience one knows all the
logical
consequences one knows It is
of anything a

controv
fear of the
system
This is all
idealised version
for an
knowledge
of
A logic of belief
is more problematic than one

are not entirely consistent

for knowledge as we

in our belief sets Still Hintilika

Suggests

x̅ Bap Baba
is true He argues that the two formulas
Bap and Bq 20 7Ba are inconsistent On the
other hand he suggests none the are
of following
true Baba Bap
Baby BP Bank.BR

The semantics Hintikka is in terms what he calls

of of
model sets and his
proof procedure is
what has since become known as tableau
a
precursor of
rules On the
Usual semantics
consists an
for classical
logic a model
of assignment
of
a truth value
to
every propositional symbol and an
interp
of the connectives boolean
logical as
functions
on these truth values so that each
complex
sentence
of the language is assured a
unique truth
value Hintikka contrast
by takes a model to
be a model set a collection
of sentences
deemed to be true which
might be
thought
as a
partial description of
of
the world

A model governed
set µ is
by
the rules Hintikka set rules
following

C Pem
If then
7 M

c n
If PnQ EM then PEM QEM

C V
If P VQ EM then Pem QEM

777 EM
If
c n
then PEM

c n
If Rna EM then spear Qem

c.TV RvQ
Ifn eM then Penn Q EM

These are
just the classical conditions

for mod
log we need the notion
of
a model system a collection
model sets an alternativeness
of
relation
one for each knower a
The following additions govern
the modal
ops

C 7 Pem
If and
if µ

to a model
belongs system
oh then there is in or
at least one alternative

Mt to µ with respect to

a such that M

C KK
If JtaQ EM and
if n
is alternative
any to me

w
resp to a in model

system or then Kalent

 C k Karen PG
If then M

c 1
If Ha em
thenfa
P PEM
If7PaPeM CCKate the
µ

The procedure is
proof big
reductio the
if assumption
that X
belongs to a model
set leads to an
inconsistency
then must
to
belong every
model set i e it must be true
n all models which is to
say
that it is valid
e.g show
KapakaQIIK.at 1Q

Ka rKaQ ele Assumption

2 7
2 0 EM Assumption
for reducho

3
Pan 2nd EM C 7k 2

4 PnQ EM C.pt 3

5 HAREM
6 KaQEM
3cal 2

7 Na Peut C KK 5

8 KaQEn C Kk 6

9 PE u c k 7
W Q Emd c k 8
11 PEM 1 1
12 Q EMA x1
In 14 µ is an alternative model set to µ
Mt and fin are the two alternatives
deriving
4.1 one is impossible be
from of 9 4

the other impossible be of 10 12

be
Epistemic logics continue to
of
interest to philosophers Nozick 1981

has raised some interesting problems with

Kac 0Q D Ka KaQ in regard

to skepticism Let E be some
experiment
statement I see a human hand
e.g
me Let S
before be the crucial

statement that I unable to

skeptical am

disprove e.g I am
dreaming or alternative

I brain in It clear
wax is
quite
am a a

that E is true I see human

if a real
 hand before me and not smt that just
appears like one then S must

be false i.e it cannot be that I

am dreaming have
presently we

1
E 0 5

and since we know this to be the case

1
KalE 0 5

Now the skeptical position is

just that I
rule out the
am unable to
possibility
that I am dreaming or am a brain in a Vat

111
Hans
From we have

11111
KaCEo 5 0 RAE 0 a S

contraposing the consequent we have

A KaCE 2 5 7 Lars D KaE
from 11 and 111
applying MP twice

if Ka Eons then Ka 75 Na E

Ka E 075

therefore Hans D Fu E MPI

États then 7
taE
THATS
therefore a E mp3

We derive K E

Note that
for
P 0 220 P3 Pn Put

If we know that Pn then

Applying Modus Ponens n times we
get
Puti
 In I have no experimental knowledge
effect
Nozick 1981
that we should reject
suggests

but on the other hand could

be correct This shows us how
logical
formalisation helps us to scrutinise philo

puzzles

Exercise

Prove x̅
using Hintikkies model sets
1

I JE.K.PT Ya

If by reductio

1 Kaks µ Assumption

7ha EM Assumption
3 87 Em Conk 2

4 79 EM C 2 7 3

5 Rak Peut C Kkk I

6 KREM C K 5

7 PEN C
11,6

i K.tn JKaP
2 Prove x̅

7,2 0Pa
Karen assumption

Papen assumption

KanPEM c 72 2

PEM c kt

REM C K

2.208
3 Pru x̅

KaP Kaka
Ka Kaka Kaka Ka
I 2

1 1 Ka EM assumption
2 Kakadem assumption
3 Pankapen link 2

4 742 end C2 7,3

5 Kap Ent C Kk

2 1 Kaka EM assumption
2 7KqPe
µ assumption

3.4796M Cit 2

4 72 Eat C 3

5 KaKPeu C kk

6 Kapen C 11,5

7 Pent ck 6

R Rak
historical highlights

Armed with can at some

formal machinery we glance
historical problems involv mod reasoning

Aristotle's Development the

of square
99.33
Aristotle's future sea Battle
Pg 35

Ancient Greek Philosophers were much concerned

w relation beto and
necessity determinism
Events do not occur
haphazardly but rather
with carrier
regularly events the
affecting
course
of later events

its roots in an earlier

Ifevent an

A
event B has
then it
wouldseem that
gift B could not
help
but occur
and the earlier
A could
if
not be altered
event

then it would
be her so we have not
 2

the relativenecessitt 13
of given A but
the absoluteneeessity
ofB.Everytn.mg would

then seem to be occurring

of necessity

Aristotle meant as the

by necessity
unalterability
of whatever has already happened Ackrill 1963,9113
Now unalterability appears to identify a

fundamental asymmetry beth past future the

past is unalterable in a the
way
not In
future is one
of the mos

famous arguments found in De

Interpretation 9 Aristotle
suppressed
issues and faced the problem
of causation
of determinism in terms
of truth
truth at a
given time and necessity
One cannot undo what has
already
been done But it is true
if not
that be battle tmr
there will a sea

the also determined and

future appears
unalterable and the
asymmetry
See
collapses Ackrill 19,63 p 50 51

This is the
famous problem
of
future contingents Aristotle
clearly
rejects the
fatalistic or
determine
conclusion
Absurdities he calls
them
if everything is and
happens would
of necessity there
be no need to deliberate For
his solution
offered see Ackill 1963 ps2 s
Aristotle denies that events in the

future are
already determined
his response to the 1
argument to

deny that statements

about the
future are in some
sense
true or There
already
false is
considerable
disagreement amongst commentators as
to what he
meant by the's The
next has subtle
complexities
us showing
a
modal story

Aristotle claims everything necessarily is

is not It unclear he meant

or is
if
PvP or Rv Did Acknit claims

that Aristotle was

clearly arguing for
 that this
the former Aristotle emphasises
does not
nterpretation place special
any
the
statements abt on
premium on
future
a serond reading Aiki states Aristotle

holds that a statement

w a
truth

value has a necessity value

automatically
true necessary but he
if if false impossible

claims that a statement may lack a truth

and may acquire one later pino

val

create and Kneale 1962

favour
a
diff
reading for them Aristotle holds that
there either will
although Item
won't me
be or be a sea battle one

that either it is true that there

annot infer
will be a
SB tmr of it is true that
there won't be a sea battle ther

in
See discussion of Supervaluations

Franssen 1966 and for notion definite truth

of

I
show p is not valid in le

need to show
Counterexample 422130
show 17 Apr 711 7139

xp here 2 as that
would imply Alte n wit

11 SIN p but sina.ae

P It APs P not always true valid ink

The Master Argument of Diodorus crones

Two schools distinguished in

of logic
were

antiquity the Peripatetic school of

Aristotle's
teachings and the Stoic School founded
 f
Zino The most famous
logical
y
the stoics was Chrysippus
practitioner of
who was influenced by the Megarians

biodorus crows Quito The Master

is built the following
argument off
tarting points there is an
incompatibility

betw

1 Everything that is past true is

necessary

2 The impossible does not follow

from the
possible

3 what neither is nor will be is

possible
Seeing this incomp Diodorus used

the of the first 2 to

convincingness

establish the thesis that nothing is

possiblewhiinwitheri.sn illbe
true

that loud
Aristotle believed Smt

happen even though it never does

938 for more

The one and future conditional

some modern logicians read

if then
q as the material

conditional i e it is false if the

antecedent is true and consequent
false otherwise it is true Some logicians
have called it the
shifeanonditional
after Stoic Philo There
logician
are
diff interps of the material
conditional Read also about the
Paradoxes

Heredity These are

that a statement
false materially
implies anything i.es 0 90Q and
2
a true statement is materially implied

by anything
0
p Q Lewis 19181

sought to Stricken the connec beta 280

and introduced 8
P Q expresses strictly implies Q

Q 20 Q We term
may
this the diodorean conditional for
more reading see 241

The Necessity
Reality of
To understand the
controversy of
mod tog
we must
investigate
the modal is part of
if the

content judgement
of
a

The content a
cognitive of
Sentence the proposition it is
expresses
 by its
commonly thought as encapsulated
of
ruth conditions the the
it purports
features
to
of
reality describe may
make it true

The prop that John is a man

is true the individual John has

the
property of being a man

But the
prop that John is
necessarily
a man will
require more
Fitz
truth then that the individuc
just
John has the
property being a
of
man also be
there will have to

Sunt to his
in reality corresponding
necessarty being a man That

smt.no is
usually thought of

as a real essence a
property
an
individual has necessarily
i e
property that is his
a
part
of
nature so that
essential he

couldnotexistifhelackedn ky ua

essencer natural kinds natural

related notions were

laws and

derived from Aristotelian metaphysics

This Prevalent way thinking

 admitted necessary connections in

beth individual and

reality an

some its properties real

of
essences beta a
group
and some
of
individuals their
of
common properties natural kinds

between an individual or
grop
doing something and another
poss

the same individual or

group's
doing suit natural laws from
mode

t.fi
much
fffeita
the same
conceived

way as
of in

any
 other
part
of proposition
a

namely as
representing
a

feature of reality that

determines the truth the

of
proposition

understand how
one can
readily
a healthy skepticism about the

necessities in nature will

nudge
one to the Kant Frege position
which denies that models
belong
to the content Claim The
of
a

treatment of causality in
Hume 1888 captures this view

A proposition that events

saying

of type A cause events

of
type B is
widely thought to

make a claim about the

way things
must be Whenever events
of
type A occur events
of type
B must occur Hume however

denies that there are necessary

nature and this holds
conditions in
if
then proposition can
no
part of a

represent them as anything that

determines the truth the prop
of
So the must in the causal claim

is not part the prop at all

of
Its role is elsewhere
assigned

for more reading on

diff views

See 943

Given the modal inference rule

characteristic
of Necessitation

nomad model
of logics

if 1
then 1 2
 symbolises it is asserted

that The axiom

of
necessity is
T K1
P 02 holds in not

and with both these

of
we obtain the desired rule

which holds in K

1 P 2

This makes it look as though

there really is no
logical
in content bet
 assertori and apodeictic
judgements But ofc this is

not so P and
P
differ in content

must be
distinguished
the
as true from usually
false 2 132 The
collapse of pre modality

effected by p 2 is

avoided although Pop

is true its converse is
 not 20133
usually

The realist about necessity

has no
problem taking
a

modal as
part the
of
The antirealist
proposition
about
necessity takes the
modal out fire prop and
of
it up a level either
raises
a comment abt the en
as position

or as
representation noncognitively

an about
expression one's attitude
of
 the propositional representation
u2itabk.am roogSystems

a is an argument that
Informally proof
convinces formally a is
proof of a

formula and is a
finite object
constructed according to fixed
rules that refer to
syntactic only
and not
the structure of formulas
to their intended meaning

The syntactic rules that define proofs

Specify a
proof procedure These are

sound
for a particular
logic if any
formula that has a
proof must be a

valid formula
of the
logic It is
 is valid
Gete for a
logic every forma

has a
proof A sound complete proof proc

allows us to produce witnesses

props
that valid
formulas are

There are
many kinds proof procs
of
Loosely the 2 broad
categories are

synthetic analytic
An
analytic proof proc
decomposes the
formula into
being proved simpler simpler
parts A builds
synthetic proof proc
its way the
up to
formula being
proved
The most common example of a
synth
Proof proc is an axiom Certain
system
formulas are taken as axioms A Starts
pf
w these axioms and using rules of
inference produces new formulas

Tableau Systems are

analyt pf procs and
are
ref orsystms.TO Rv a
formula
we it
begin
by negating 5 contradiction

Different forms these it

exist but
of
we

focus on
grefixedtabu.rs

It is easier to discover proofs in

tableau systems than in axiom

systems
 Tableaus

Proofs are trees prefixed formulas

of
which are used Trees
by prefixed tableaus
are a kind
of graph

IF
I

Node each box prefixed formulas

Zoot Node A
Leaves F G H
Children directly linked nodes to a parent node
Parent directly linked node to children nodes

Path Any sequence

of linked nodes from Root node

Branch A maximal
path from root node to a
final node e g F

Definition Prefix
A prefix is a finite sequence of the integers A

prefixed formula is an expression in the form

where 0 is and
prefix
a
X is a
formula
We will write prefixes using to separate
fullstops
integers The idea is that 6
hates a possible
world in
2 some model and oX tells

us is true at the word

6 names
5 n where ne It should
always name
a word that is accessible from one hat
5 names
Definition Branch Extension Roles

Conjunctive

nY on Xvi 404
F I 3

Disjunctive

Double Negation

Possibility on new to branch

Mix
III
r

a nnX
Basic Necessity 5 n
may alreadybe on branch
on

Definition Closure

A tableau branch is closed if it contains

both oX and r X A tableau is closed

closed
if every branch is A closed
branch represents an impossible state
of
affairs

Definition Tableau Proof

A closed tableau for 1 Z is a tableau

pf
Z and Z is a theorem it has a tabet
Ey Tableau
f of
PnQ 0 Rn Q

1 NQ 0 21110
1 AQ
1 21130

1 121,0
closed sea

Exercise
1 for k Qu 21 Q NQ

1 n 139 40 712101
1 91479
1 7 PAQ
1
1 QQ
 i.im iat

1 17 1.170
1 1

2 For K Qu AND a Rna

1 In a 9107

1 Pn Q
1 7 41210
1 9
1 Q
101 IP
Q
i 7419 9

iii.int

t.mn
8
Note DX has a tableau proof then so
if
does x is a thm so is
if
More Tableau Systems

We have air seen a tableau

single
system which proves exactly the valid

formulas K
of
Now we add various rules to that
System to prod for
Pf precs
the other mod
logs
Definition Special Necessity
for 5 and G u
already on branch

ox I
 on x
B x

4 T
4

Rules
Logic
D

T T
114 4
134 B 4

54 714
 SS 7,4 Yr

whilst there is
nothing wrong
with this 55 tab Sys there
is a
simpler version Take 95

Prefixes jst next not

segs
them
of replace
all the

modal rules w
following

Definition 55
Possibility knew to
branch

MAX
KX
 55 Necessity ie air on

Definition branch

Let us conclude this sec with

an
example using this simplified
55 trans
system ref symm

Rf Rv 2

1 2 R
1 an R
l p
l 2
 2 n P a

3 77 p

3 P
2 P

Consequence Tableaus
logical

Previously we discussed
logical
consequence
in a modal setting
is
SKU X which more complex

than CL be and local

Of global
assumptions Recon that X is
true at each world model
of
a

at which the member U

 f
are on true provided to

members s
of are true at

EVERY word hat mode

of
Tableau rules his are simple
for
Definition Assumption Rules

Let L be one the mod

of
which rules
logs for tableau
have been Let S and U
given
be sets formulas A tableau
of
uses
sfglobelassum.ph and
 uaslocclassumptions
i.fm
following 2 tableau rules are

admitted

Local Assumption Rule Y

if
is
U then
any member
of
14 can be
added to the
end
of any oger branch

Global Assumption Rule Y

if is
member
any 5 Y
be added to
of then can

end
of any one
 brnchonwhichtappeasa

epx TL
assumptions are involved in

a
tab construction for 1

the tableau is
derivation
a

instead of
of a
proof

earlier we proved 2023 k and

holds now a k derivation

give
2052 using 209 as
of
8 G